FIGURE 1.

Schematic representation of the workflow followed to predict K. pneumoniae-targeted host proteins and subsequent downstream functional analyses with the K. pneumoniae predictome. To predict K. pneumoniae-targeted host proteins, we devised a three-step computational strategy. In the first step using HPIDb (an exhaustive repository of experimentally determined host–pathogen interaction datasets), we identified potential interologs between K. pneumoniae and humans at e-value 10^{– 10}, 90% query coverage, and 50% sequence identity. Further, the K. pneumoniae proteins predicted to interact with host proteins were screened based on their fitness in mouse (Fitness data derived from Bachman et al., 2015). Those K. pneumoniae proteins with a fitness value > 2.0 were retained in step 3; an in silico GO term similarity check along with secretion propensity checks was carried out. The final KHPPI thus filtered were subjected to further downstream network analysis.